My name is Ruplu Bhattacharya, and I'm with the IT Hardware and Electronics Manufacturing Services Equity Research team at Bank of America. Today, we're honored to have Jon Faust, the CFO of Sanmina Corporation. Jon has been with Sanmina since 2023, but he has over 20 years of experience with another company that we cover, HP, and various branches of HP. He's got lots of industry experience, and we've got lots to talk about. Jon, thanks for coming.
Yeah, thanks for having me, Ruplu. Good to see you, as always.
I want to start with something that's on everybody's minds, and that is ZT Systems. If I remember from the last earnings call, you showed us a slide that showed the evolution of ZT Systems in three stages. Can you remind us of that? Where does ZT stand right now in terms of qualification and testing, and how do you see its evolution progressing over the next quarter, two quarters?
Yes, absolutely. Whenever you do a deal like the deal with ZT or an acquisition, it's very important to have a well-thought-out strategy or just a plan, effectively, of how you make the investment thesis successful, right? What you want to do with the business exactly. To your point, what we tried to do on our last earnings call is lay out that plan in very simplistic terms, very high level terms, just to give investors and sell-side, buy-side, a better idea of where we were at in the process. When we first closed the deal at the end of October last year, we just wanted to do the initial integration work and so forth.
What we were also doing is taking a bit of a risk, continuing to make investments in the business, because we knew that there were certain elements of ZT that were going to transition over time. ZT was making a lot of investments when they were a standalone company, again, when they were part of AMD, and we wanted to continue with that. That was the first phase, the first things that we needed to do. The second phase was all about winning business. Maintaining the business that ZT had across all three lines of business. That's general purpose compute, CPU-based platforms, storage, and then also accelerated compute, the GPU-based platforms. We had to go win that business. As we announced on our last earnings call, we have won that business, both pre-production and production stage business, which is great.
To the question that you're asking, what we're working on right now is, we're working through the pre-production phase, and we're starting to lock in what production will look like. We've got a pretty good idea from the customers already what the forecast is going to look like when we get into production at the end of this calendar year and into next year, but there's still a lot that we have to learn in pre-production. Everything's very much on track, and we're very excited about it. When I think back to that plan that we set, we've been executing very well to that, everything that we've said since day one, we just want to continue to do that. Right now the stage is do the pre-production testing. We're getting units to AMD, for example, to customers and so forth.
We're learning a lot on how to produce and manufacture the products, then that will help to finalize the eventual production schedule. That's when we'll come out with more details from a guidance perspective, like the scale, the revenue, the profitability of the business. Right now, we're right where we expected to be, in that overall plan that I mentioned.
Got it. You mentioned different buckets and different things that ZT does. Can you elaborate a little bit? Talk to us about what ZT does in terms of accelerated racks, non-accelerated racks, and how should investors think about overall long-term trends in these businesses and revenue growth?
Yeah. ZT, historically, the company's been around for 30 years. I'd say the last 15 years, they were focused on the data center business and as an ODM player. They were a true ODM, competing with the large Taiwanese ODMs, for example, across those three lines of business that I mentioned. CPU servers, storage, and accelerated compute, they were fantastic at it. Now, those are all lines of business that we wanted to get into Sanmina, because if you think about the core Sanmina business or the legacy business, when it comes to what we were calling cloud infrastructure, we were predominantly focused on data center networking. This ZT deal helped to expand our TAM. It got us back into server integration work, storage. When I say server, both CPU and GPU business.
We are excited to expand into that, and we want to maintain that business. It was a true expansion of our TAM at the end of the day because Sanmina had gotten out of those businesses. What ZT always was focused on as an ODM was just the final system integration and test. Really like L10 to L12 manufacturing. They didn't do any sub-assemblies or anything below that level. That's what Sanmina brings to the table, and that's what makes this deal and this transaction synergistic for us. We want to win business where ZT can continue to do what they've done very well in the past, but then complement it with what Sanmina can do. We can do board fabrication, we can do rack fabrication, a lot of things that ZT wasn't doing.
First things first, back to the plan that I mentioned about achieving the investment thesis. It was winning business, maintaining the business ZT had with their legacy customers, win more customers with what they do well, and then over time, introduced some of the capabilities that Sanmina brings to the table with vertical integration. To answer your question directly, maintain, stay in all three lines of business.
Got it. Can you remind investors what you've telegraphed in terms of revenues for ZT for this fiscal year? How much revenue have they generated so far, and what should investors be thinking of for the rest of the year? If you can segment that between AI related and non-GPU related.
Sure. We announced the deal May 19th of last year, almost a year ago exactly. At the time, we said a couple of things. One, we said when we expected to close the deal, which happened at the end of October that same year, we expected the revenue run rate for total ZT to be in the $5 billion-$6 billion range. That's held true. If you think about the full year, fiscal year 2026 guide that we gave in our last earnings call, the ZT guide has you right in the middle of that. Now, of course, there's a range, it could be towards the higher end of the range or a little bit below the midpoint, but solidly within that $5 billion-$6 billion.
We had mentioned that all the way back on May 19th before we even closed the deal, and this is why I say, Ruplu, earlier that we're very much on track on our execution of our plan for ZT. What we also said back on May 19th a year ago was that we expected to double Sanmina, the size of the company, speaking about revenue, within three years. At the time, of course, Sanmina was about $8 billion, so that applied a $16 billion number, but we said within three years. Fast-forward to when we closed the deal, we accelerated that and said, "Hey, we expect to do that within two years." We brought that forward by a year, which implies our fiscal year 2027, which is coming up soon. On our most recent earnings call, we said $16 billion+.
Now that we've got customers lined up, we've got a better point of view on the production schedule. Still preliminary, albeit, so not really a formal guide, but we said $16 billion+. Jure made the comment on the call about becoming increasingly confident about the business. That's because we had a plan and we've been executing one step at a time. It's been going well, and we just want to continue to execute to that plan.
Got it. That makes sense. I'm going to come back to the revenue streams in a bit, but let's talk about manufacturing facilities. I think you bought three facilities when you bought ZT Systems. Can you give us an overview of how those facilities are trending? Are they set for building data center racks? Do you have enough power? How should we think about the production capacity of these facilities?
Yeah. It was three facilities, three plants that we acquired as a part of the ZT deal. The primary headquarter location plant is in Secaucus, New Jersey, just outside of Manhattan. There's another facility here in the States, in Georgetown, Texas, just north of Austin, and then a third one in the Netherlands. All of them were capable in producing business already across those three lines of business that ZT was in, and state-of-the-art facilities. Now what we've been doing, and this started even before we did the deal, and again over the past couple of years, even before we closed, is there's been investments made in those facilities to retrofit them effectively, for the next generation of accelerated compute, because the power demand is incrementally massively larger.
That's one thing as we were talking with customers, they want to understand the power roadmap that you have, and you need to have those agreements in place with the local governments, the cities, municipalities and such, and we've got that. We had those agreements in place, and that's what was really attractive to us about the assets that we picked up as a part of this deal, is that you not only had plants executing at scale across those three lines of business, but for the long term, you had agreements in place for power. You also have to retrofit the test cells for liquid cooling, so you have to have large chillers in the facilities and so forth. We've done that. We had to keep that pace, and that was going to be important to win the next generation of business.
Those investments have largely been made now. There's more money to spend over time as power comes in, but we've retrofitted that, so we're prepared. That was critically important. Now as we're in the pre-production phase of the next generation, we're already leveraging those capabilities to build these products, and we're learning more every single day in the pre-production phase. I'm talking engineering validation, design validation, and each customer has a slightly different schedule, different configurations of the product. You learn a lot, and that's important in our business, is to be involved in that stage to make sure you really learn what it takes to build these products.
Got it. You've guided for strong revenue growth, $16 billion + by fiscal 2027. The question that I keep getting from clients is, do you have enough capacity to meet that demand? Hopefully as you grow with AMD, you get more rack orders. Do you have enough facilities to satisfy the demand for the revenue growth that you're going to see?
We do. We've definitely got the capacity now to do the $16 billion+. We'll learn more over time. We've got estimates right now as we've been learning through that pre-production phase on what it takes to build the product. Because then you start to estimate how many racks can you build in a given week, month, quarter. It all depends. Right now, when you're just doing pre-production, not the full-scale unit production level silicon and so forth, you've got estimates on how long, not just to integrate it, but to test it, because you got to ship it as a complete turnkey product to the customers to put into their data centers. We'll learn more about that, but we've got the capacity in place. Now, longer term, Sanmina has a very large footprint.
Now, not all of our facilities around the world have been retrofitted, have the power or the liquid cooling test cells, for example, to do accelerated compute. we certainly do for storage, for example, or CPU-based, general purpose compute, pick your term there. we've got different levers that we could pull essentially if we wanted to move business. if we're very successful with accelerated compute with customers and we need more capacity, we could leverage the ZT facilities that are specialized for that, and then potentially move other business elsewhere or bring up new plants.
Got it. You mentioned this non-GPU, non-accelerated business. How do you think about that market, and what are the pros and cons of focusing on that business?
The news over the last, what is it, three, four months, whether it's from AMD or I know HPE, prior company of mine, announced recently this week, general purpose servers, CPU-based platforms are doing extremely well right now, and that is one of the lines of business that ZT is in. Now, for us, as a part of that plan that I mentioned earlier, we were very focused on making sure the transition of accelerated compute was successful. That doesn't mean that we're ignoring the CPU-based business. You just have to make sure that you don't have too many priorities at one time, because if you start to slip with these customers, especially hyperscalers that want product at scale, they don't want the issues. You don't want to get yourself over-committed. That is interesting business to us.
Our strategy and approach as a part of that plan was maintain the legacy customers that ZT had with the programs they had, convert to the next generation of accelerated compute, and then expand from there. Now, we've talked a lot externally about vertical integration, doing more things with the rack because of Sanmina's capabilities. Like below L10, for example. It is an expand, like the program set that we have, too. The CPU servers, and even storage products. We are interested in that, and we do have a team in place that focuses on that. You got to take one step at a time, and then you can win more programs, especially with the large customers.
Got it. Maybe I want to transition into margins now. Talk to us about how you see ZT Systems margins trending over time, and if you can address two things. One is consignment versus not consignment. How does that impact things? Just in terms of the AI revenues versus non-accelerated revenues, how does that mix impact margins?
Yeah. The operating model that the customers choose, so consignment or not consignment, which really just means that we never take financial control of certain components, which means in our business, we don't recognize revenue on it. Like that's all dependent on customer. You do see it across all three lines of those businesses. Some customers very much prefer like a buy-sell model where they control certain components, whether it's the CPUs, the GPUs, memory, et cetera, and then we wouldn't rev rec that. We're fine to work in any of those models. For me, from a CFO perspective or a financial perspective, the true economics is how much profit are you making per rack or per unit that you're building and making sure that the ROIC is good on that front.
Now, in the short term, with the mix of the business that we have right now, we've said that the margin profile is similar to core Sanmina, and you've seen that play out. It's actually been a little bit more favorable than that in our Q1 and in Q2 results and even our guide for the back half. Core Sanmina exited fiscal year 2025 at a 6% operating margin. We're already there as core Sanmina. If you look at Q1 and Q2, we've done better than that. A little bit better in Q1, quite a bit better in Q2, and our guide for the second half is even better. ZT's done well from a margin perspective, but the core Sanmina business is continuing to make progress as well. That's our best estimate for right now.
Now, we haven't formally guided FY 2027 yet because we have to lock down the production schedule with the customers. Also, the final mix of the customers, that'll dictate the operating model, so that could change the profile. Our best guess right now, and we've been saying this since day one when we quoted the $16 billion for revenue is, there'll be a good mix of consignment in there, similar profiles for now for around that Sanmina profile, think like 6%. Over time, if we're successful with vertical integration, that wouldn't be more revenue throughput, that would just be profit dollars, so that should drive accretion.
Got it. Last quarter came in better than expected for ZT. I think also ZT used to do some legacy accelerators as well, both for NVIDIA as well as the earlier versions of AMD's accelerators. Talk to us about that. Is that a business that you would like to be still in, and how do you see that trending?
Yeah. in Q2, we didn't have almost any, I think it was close to zero, like legacy accelerated compute platform business, so NVIDIA-based business. That had gone to zero, which we always knew was going to happen because of the nature of the transaction that AMD did. What we did have that we were very pleased with was current generation, new business for us, like AMD platform business for accelerated compute, and we executed very well on that front and got some incremental. There was strong demand, too. Now, we always expected that to be batchy business for us. The real opportunity for us is the next generation of accelerated compute. the reason being is by the time we did the ZT deal, manufacturing partners for the current generation of accelerated compute was already done.
The dynamic that we saw in Q2 was some customers wanted to test us out, test out ZT. Can we build a different platform of product? Learn how to do business. That was a very important precursor to winning the next generation of accelerated compute. Prove that we can do it today. When we had guided the full year, we had some of that demand in place. We expected it to happen between Q2, Q3, maybe even a little bit before in Q4. That was because of when we thought certain components and supply was going to become available. Once more supply became available, this is the strong execution point. The customers wanted that product immediately, and ZT did a fantastic job building that product, shipping it. You saw the results that we had, which was essentially an acceleration. Our full-year guidance didn't change.
Our outlook's still consistent, but the timing changed because of the strong execution, and we got some more demand, too. We'll see if some more comes in, but right now we're really laser-focused on getting prepared for the next generation to make sure that that goes well.
Got it. Building racks is a working capital-intensive business, right? As you build more and more racks into inventory, how should investors think about working capital, inventory and free cash flow?
Yeah. You're absolutely right. Over time, our working capital will build. The dynamic that we've seen in the first two quarters with ZT, though, is that legacy business kind of coming down, right? Because we knew that dynamic was going to be happening, so we've been generating a lot of cash, which is great, and we've been preserving that cash. If you look at core Sanmina, we used to always hold somewhere around $600 million per cash in any given quarter when we would exit. Now we're $1.5 billion, in that range. We've been generating a lot, and that is to get prepared for the eventual working capital build. I expect that to happen towards the end of the calendar year. Some of it will start here, even in Q3 and into Q4.
depending on the operating model, like how much is consignment versus not, because if it's consignment, then we don't have to go pay for those materials. We receive them physically, so that'll play in. I do expect working capital to build. what I've talked about externally is I think about it from a leverage ratio perspective, and Sanmina had the best balance sheet by far in the industry prior to doing this deal. even post doing the deal, we're right in the range of our peers and competitors. I set a target range of 1-2x net, and we've been well below that so far. I do think we'll come into that range as we get into the back half of the year. depending on the operating model, could even go a little bit above it, say into next year. We'll see.
It depends on that final production schedule. that's all been part of a plan. That's why we set up the capital structure that we did, not only to purchase the assets, but to get prepared and to have the balance sheet and dry powder to support the growth of the business.
Got it. Okay. That makes sense. I do want to touch the rest of the business, but before we get to that, are there any risk factors to ZT revenues for this year that investors should keep in mind?
For this fiscal year, there's always risk, right? If I think about the ZT, the big risk factors, one of the big ones was were we going to win the business? Were we going to stay in the accelerated compute business, and win the next generation? That's what we talked about in the last earnings call, that we've essentially kind of crossed the chasm from that regard. There's still a lot of work to be done, right? We're in this pre-production phase, and we need to execute. The biggest risk right now is not executing, and that's why we're so focused on let's make sure that we execute well.
When customers want certain pre-production racks, whether it's EVT engineering validation or design validation, whatever component makes or configuration change that they want, we need to execute well on that, because that's really what we're doing this fiscal year for us, right?
Okay.
As we get into FY 2027, that's when the production scale will happen. We've got to get prepared for that. For this fiscal year, it's really just kind of executing well into pre-production, continuing to win customers, kind of lock down that demand. Same thing on our core business. That's normal for manufacturing.
Got it.
You have to execute. Otherwise, customers will go elsewhere.
Let's move on to the rest of the business, because if we think about legacy Sanmina, you have a very good footprint in optical communications. talk to us about how the communications market is trending, and how do you see Sanmina positioned in this?
Yeah. The segment, or what we call it, is communication networks and cloud infrastructure. Just talking core Sanmina, legacy Sanmina, that part of the business has done extremely well. It's seven quarters in a row now that we've been growing around 20%, 20%+ , and that includes the optical part of the business. The reality is that we could be doing even better there. There's a lot of demand, but it's constrained, right? On the cloud infrastructure side, whether if you think data center networking, it's different chips and so forth, memory, things that come into play. On the optical side, it's lasers that you need to do the manufacturing. The demand is there, and we expect that to continue. The core business overall, and that's been a big reason why last fiscal year, the core business grew high single digits.
We've guided this year that'll be the same. It's a little too early to say. We haven't guided FY27, but on our last earnings call, we said that we don't expect that dynamic to change anytime soon. We'll see what happens with the supply environment. If it stays constrained, similar type of profile. When that eases up, we think we could do even better. On top of that, Ruplu, we're always looking to win incremental business, take business, a larger share from customers. Hopefully, we can do that, too.
Yeah. I want to address some other end markets, but before I do that, maybe I'm going to jump to the capital allocation question. As you see working capital needs going up, how should we think about your allocation of capital to CapEx? Any M&A that could be relevant, and how should investors think about returns?
Yeah. capital allocation strategy for us has been very consistent. It's all about driving growth, right? That's what we want to do at the end of the day. Now, we always look at it from an ROI perspective, so I'm always balancing the ROI of paying down debt, doing share repurchases, things of that nature. we're heavily indexed to how do we preserve the dry powder to drive growth. that's why you've seen our cash balance grow over time, because we know that we're going to need some of that cash to drive the growth, depending on the operating model of the future. that doesn't mean that we don't continue the pipeline going from an M&A perspective.
We're always looking for opportunities to grow and expand our business, and we might do different acquisitions on that front, whether it's bringing in engineering talent or bringing in new capabilities, automation, things of that nature, not just for ZT, but for the core business too. That's why it's important that we have this strong balance sheet. The balance to that is I'm always looking at that leverage ratio and making sure that we stay within the right range and we don't get over-levered by any means. We'll continue to do that, and we'll be opportunistic. If you go back to after our Q1 results, the stock was down quite a bit based on the market reaction.
We did do some share repurchases to offset the dilution of the shares that we granted to AMD as a part of the ZT deal, but that was because the ROI was just so attractive not to pass that up. Now it's all about driving growth.
Got it. Sanmina also has another unique business, which is the defense business, and SCI has been a very steady contributor to revenues and cash flows. What's the long-term plan for that business, and how is that business trending?
Aerospace and defense has been a great business for us, and we've got a mix of traditional EMS contract manufacturing work, but we do have our own products and IP in that business, and it has been growing, and it's an accretive to business if you look at the overall profile. It's definitely something that we want to grow and invest in. Based on all the conflict in the world, which is unfortunate, there's no shortage of demand there. There's new players coming into the market, not just the legacy primes that everybody knows about, the large players, but new companies that are coming in too that we've been targeting. More to come when we guide FY 2027, but that has been an area of focus for us, putting capacity in place, winning new programs with new customers.
We think that that'll be an attractive end market for us. Maybe not quite the same scale as AI these days, but it's a good business to be in.
Got it. How about the industrial end markets? What's happening in there?
We're starting to see that come back for us. If you go back a couple of years, all the end markets that we were playing in were working through inventory absorption, right? A lot in the channel, and so forth. Industrial was kind of similar. Whether you think about semi-cap equipment, that's starting to get better, so the inventory profile is getting better. If you look at our core Sanmina business, when we talk about the different segments, that was in Q1, industrial, medical, aerospace, and defense, all those groups, it was a bit of a headwind, down like 3% year-over-year. It was flat in Q2.
As you look at the back half, it's going to return back to growth, and part of that is the improvements that we're seeing in the industrial space, and that's across both semi-cap equipment, but then other areas too. think like police handsets, radio handsets, cameras, equipment like that for security purposes. That's been doing quite well already, but we expect that to get better. Again, maybe not the same growth profile of some other end markets, but still a good business to be in.
Got it. We've got about two minutes left, and I want you to leave investors with what you think the market is missing about Sanmina's story. One question I keep getting from clients, I want you to address this as well. The AI rack building space is a very crowded space. There are lots of companies, EMS, ODM, who can do this. How do you see Sanmina's and ZT's systems positioned in this market? What are some of the unique things that ZT brings that let you compete in this market against those players?
Yeah, let me start with the ZT side first, just to explain that. Yes, it is a competitive market, but it's important to keep in mind that ZT was a Tier 1 provider in that space, competing with the Taiwanese ODM. Their manufacturing capabilities are second to none, right? If you think about the facilities, the people, and so forth, so we're very excited about that. Now, certainly, there was the transition period that we had to work through, but we've had a plan for that, and we've executed well to that plan all the way to back to what we said a year ago. Things have been happening exactly the way that we said, even better than that. We need to continue to execute on that because the growth profile of that end market, in particular, is massive.
I think that's missing in our story today, is the potential of that business, both accelerated compute and then the other two lines of business. If we execute well, and I think that we've shown over the last year that we can execute well. ZT had those capabilities, right? To your point around the crowded space. Very focused on that. More broadly, we've learned the lessons over time. Sanmina has, certainly Jure has, that it's good to be diversified, and there's a lot of opportunity in the other end markets that we play in. Communication networks, our legacy cloud infrastructure business, again, that's been growing 20 %+, and we don't see that slowing down anytime soon. Energy, we're investing in. We mentioned a couple quarters back about expanding into medium voltage transformers. That's certainly benefiting from the AI ecosystem.
Pretty much across all of our end markets. Aerospace and defense, we spoke about. Medical too. We're going to continue to invest in those, and we do have the balance sheet to do that, and we've shown that we can execute across that profile. A lot of opportunity there. We're certainly up to the challenge, but we're very excited about the future, and we think the potential to both grow the business and expand margins is there. We've been executing on that ever since I've been here, certainly too, showing quarter after quarter that we can continue to grow and expand margins while still maintaining a healthy balance sheet and cash flow.
Got it. We've covered a lot of different topics, so thanks so much for coming today and giving us all these details.
Yeah, thank you for having me.